← Knowledge LibraryChapter 6 · Metric Design

Citation SelectionCitation Probability, Rates, and Denominators

How to calculate citation performance without inflating visibility by discarding no-search or no-citation outcomes.

How to calculate citation performance without inflating visibility by discarding no-search or no-citation outcomes.

Citation probability is conditional

For a source and prompt, citation can be recorded as a binary event. Its probability depends on the query, engine, time, language, location, account state, search mode, and competing sources.

It is not a permanent property of a page or domain. The same source can be repeatedly cited for one prompt family and absent from another.

Citation rate summarizes repeated observations

A basic citation rate divides the number of runs in which a source is cited by the relevant number of runs. If a domain appears in 18 of 50 comparable responses, its citation rate is 36% under those conditions.

The calculation is simple; choosing and labeling the denominator is the difficult part.

Different denominators answer different questions

Imagine 100 runs: search activates in 50, the source is retrieved in 20, and it is cited in 10. Citation is 10% of all runs, 20% of search-enabled runs, and 50% of retrieved runs.

All three are valid

Overall visibility, post-search selection, and post-retrieval selection are not the same metric.

A high conditional rate can coexist with weak end-to-end visibility.

Preserve null outputs

Martinez warns that removing no-search or no-citation responses can inflate apparent visibility. Reports should state whether their denominator includes all prompts, search-enabled runs, retrieved runs, or only answers that already contain sources.

A citation percentage without a denominator is not an interpretable result.

Frequently asked questions

Questions about this topic

What is citation probability?+

It is the probability that a source will be visibly cited under specified query, engine, time, and contextual conditions.

How is citation rate calculated?+

Divide the number of relevant runs containing the citation by a clearly defined set of relevant runs.

Should no-search responses be excluded?+

Not from an end-to-end metric. They may be excluded for a conditional post-search rate, but both the exclusion and denominator must be explicit.

Why can a 50% citation rate still mean weak visibility?+

It may be conditional on rare retrieval. If the source is retrieved in only a small share of all runs, its overall citation rate can remain low.

Source notes

References

Sources are listed in APA 7 style. Preprints are identified as such and should not be treated as peer-reviewed findings unless separately published.

  1. Martinez, O. (2026). Optimizing visibility in generative engines: A critical survey of generative engine optimization (2023–2026) [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2607.14035
  2. Zhang, K., He, X., & Yao, J. (2026). From citation selection to citation absorption: A measurement framework for generative engine optimization across AI search platforms [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2604.25707
  3. Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., & Narasimhan, K. (2024). GEO: Generative engine optimization. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (pp. 5–16). Association for Computing Machinery. https://doi.org/10.1145/3637528.3671900